Looks like Deep Sjeng wins cct12

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

User avatar
Harvey Williamson
Posts: 2026
Joined: Sun May 25, 2008 11:12 pm
Location: Whitchurch. Shropshire, UK.
Full name: Harvey Williamson

Re: Looks like Deep Sjeng wins cct12

Post by Harvey Williamson »

Albert Silver wrote:
Damir wrote:You forgot to mention it is using 128 cores cluster. On the other hand, It would be a complete failure, if it did not win the tournament...
Considering that huge clusters have more often not won, than won, it is a very nice victory. A hearty congrats to Gian-Carlo, and to Peter Skinner for a great tourney once again!
Sjeng and Rybka are the only top programs that, currently, use Clusters and would have good chances of winning without using a Cluster. Rybka has won CCT the last few years not sure how many were on its Cluster.

I wish we could afford 80+ i7 cores :)

A nice tournament involving MI5 and Mossad. Congrats to Sjeng and well done Peter. The only thing that spoilt it for me was the incident in the last round when Junior was forced to forfeit its game.
User avatar
michiguel
Posts: 6401
Joined: Thu Mar 09, 2006 8:30 pm
Location: Chicago, Illinois, USA

Re: Looks like Deep Sjeng wins cct12

Post by michiguel »

Harvey Williamson wrote:
Albert Silver wrote:
Damir wrote:You forgot to mention it is using 128 cores cluster. On the other hand, It would be a complete failure, if it did not win the tournament...
Considering that huge clusters have more often not won, than won, it is a very nice victory. A hearty congrats to Gian-Carlo, and to Peter Skinner for a great tourney once again!
Sjeng and Rybka are the only top programs that, currently, use Clusters and would have good chances of winning without using a Cluster. Rybka has won CCT the last few years not sure how many were on its Cluster.

I wish we could afford 80+ i7 cores :)

A nice tournament involving MI5 and Mossad. Congrats to Sjeng and well done Peter. The only thing that spoilt it for me was the incident in the last round when Junior was forced to forfeit its game.
What happened? I fell asleep last round.

Miguel
User avatar
Harvey Williamson
Posts: 2026
Joined: Sun May 25, 2008 11:12 pm
Location: Whitchurch. Shropshire, UK.
Full name: Harvey Williamson

Re: Looks like Deep Sjeng wins cct12

Post by Harvey Williamson »

michiguel wrote:
Harvey Williamson wrote:
Albert Silver wrote:
Damir wrote:You forgot to mention it is using 128 cores cluster. On the other hand, It would be a complete failure, if it did not win the tournament...
Considering that huge clusters have more often not won, than won, it is a very nice victory. A hearty congrats to Gian-Carlo, and to Peter Skinner for a great tourney once again!
Sjeng and Rybka are the only top programs that, currently, use Clusters and would have good chances of winning without using a Cluster. Rybka has won CCT the last few years not sure how many were on its Cluster.

I wish we could afford 80+ i7 cores :)

A nice tournament involving MI5 and Mossad. Congrats to Sjeng and well done Peter. The only thing that spoilt it for me was the incident in the last round when Junior was forced to forfeit its game.
What happened? I fell asleep last round.

Miguel
In the final round Junior disconnected for just over the 5 minutes allowed and then returned. The Komodo operator agreed to play on. 43 minutes later when Junior had what looked like a won game the Komodo programmer turned up and claimed the game. The claim was upheld.
garybelton
Posts: 175
Joined: Fri Dec 11, 2009 9:08 pm

Re: Looks like Deep Sjeng wins cct12

Post by garybelton »

The sad thing about this hobby is when winning becomes more important than a sense of sportsmanship, isn't it?
User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 4:27 pm

Re: Looks like Deep Sjeng wins cct12

Post by Don »

Harvey Williamson wrote:
michiguel wrote:
Harvey Williamson wrote:
Albert Silver wrote:
Damir wrote:You forgot to mention it is using 128 cores cluster. On the other hand, It would be a complete failure, if it did not win the tournament...
Considering that huge clusters have more often not won, than won, it is a very nice victory. A hearty congrats to Gian-Carlo, and to Peter Skinner for a great tourney once again!
Sjeng and Rybka are the only top programs that, currently, use Clusters and would have good chances of winning without using a Cluster. Rybka has won CCT the last few years not sure how many were on its Cluster.

I wish we could afford 80+ i7 cores :)

A nice tournament involving MI5 and Mossad. Congrats to Sjeng and well done Peter. The only thing that spoilt it for me was the incident in the last round when Junior was forced to forfeit its game.
What happened? I fell asleep last round.

Miguel
In the final round Junior disconnected for just over the 5 minutes allowed and then returned. The Komodo operator agreed to play on. 43 minutes later when Junior had what looked like a won game the Komodo programmer turned up and claimed the game. The claim was upheld.
Nothing that happened was publicly reported, so how would you have any idea of what really happened?

Your version of events is wrong in every detail and you should get your facts straight before you decide to condemn others in public. Obviously, you are getting your facts from a biased source.

There is a LOT here that I have kept quiet for a good reason - I don't intend to make any accusations that cannot be absolutely proved and you should do the same.

Besides not knowing the facts, nobody should be criticized for expecting the clearly stated rules to actually be respected. Also, nobody should have to fight to get a rule enforced and it should not even be up to the players or operators which rules they choose to honor or ignore because every result affects every player, not just the two playing.

Although they game ended with a forfeit, there is an interesting side-note. Near the end of the game, a draw by repetition happened - Junior had 2 rooks vs a king and could not checkmate (a common bug that I have had myself.) I was annoyed by this too because if you are in the mode of ignoring rules they might make a case that the game should be considered a win anyway (out of good sportsmanship) and I half expected to see yet another disconnection.
User avatar
Harvey Williamson
Posts: 2026
Joined: Sun May 25, 2008 11:12 pm
Location: Whitchurch. Shropshire, UK.
Full name: Harvey Williamson

Re: Looks like Deep Sjeng wins cct12

Post by Harvey Williamson »

Don wrote:
Harvey Williamson wrote:
michiguel wrote:
Harvey Williamson wrote:
Albert Silver wrote:
Damir wrote:You forgot to mention it is using 128 cores cluster. On the other hand, It would be a complete failure, if it did not win the tournament...
Considering that huge clusters have more often not won, than won, it is a very nice victory. A hearty congrats to Gian-Carlo, and to Peter Skinner for a great tourney once again!
Sjeng and Rybka are the only top programs that, currently, use Clusters and would have good chances of winning without using a Cluster. Rybka has won CCT the last few years not sure how many were on its Cluster.

I wish we could afford 80+ i7 cores :)

A nice tournament involving MI5 and Mossad. Congrats to Sjeng and well done Peter. The only thing that spoilt it for me was the incident in the last round when Junior was forced to forfeit its game.
What happened? I fell asleep last round.

Miguel
In the final round Junior disconnected for just over the 5 minutes allowed and then returned. The Komodo operator agreed to play on. 43 minutes later when Junior had what looked like a won game the Komodo programmer turned up and claimed the game. The claim was upheld.
Nothing that happened was publicly reported, so how would you have any idea of what really happened?

Your version of events is wrong in every detail and you should get your facts straight before you decide to condemn others in public. Obviously, you are getting your facts from a biased source.

There is a LOT here that I have kept quiet for a good reason - I don't intend to make any accusations that cannot be absolutely proved and you should do the same.

Besides not knowing the facts, nobody should be criticized for expecting the clearly stated rules to actually be respected. Also, nobody should have to fight to get a rule enforced and it should not even be up to the players or operators which rules they choose to honor or ignore because every result affects every player, not just the two playing.

Although they game ended with a forfeit, there is an interesting side-note. Near the end of the game, a draw by repetition happened - Junior had 2 rooks vs a king and could not checkmate (a common bug that I have had myself.) I was annoyed by this too because if you are in the mode of ignoring rules they might make a case that the game should be considered a win anyway (out of good sportsmanship) and I half expected to see yet another disconnection.
I said what looked like a won game because in the end it was a draw because of a bug. My source of the facts is quite good and I was there when your operator played on.
User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 4:27 pm

Re: Looks like Deep Sjeng wins cct12

Post by Don »

Harvey Williamson wrote:
Don wrote:
Harvey Williamson wrote:
michiguel wrote:
Harvey Williamson wrote:
Albert Silver wrote:
Damir wrote:You forgot to mention it is using 128 cores cluster. On the other hand, It would be a complete failure, if it did not win the tournament...
Considering that huge clusters have more often not won, than won, it is a very nice victory. A hearty congrats to Gian-Carlo, and to Peter Skinner for a great tourney once again!
Sjeng and Rybka are the only top programs that, currently, use Clusters and would have good chances of winning without using a Cluster. Rybka has won CCT the last few years not sure how many were on its Cluster.

I wish we could afford 80+ i7 cores :)

A nice tournament involving MI5 and Mossad. Congrats to Sjeng and well done Peter. The only thing that spoilt it for me was the incident in the last round when Junior was forced to forfeit its game.
What happened? I fell asleep last round.

Miguel
In the final round Junior disconnected for just over the 5 minutes allowed and then returned. The Komodo operator agreed to play on. 43 minutes later when Junior had what looked like a won game the Komodo programmer turned up and claimed the game. The claim was upheld.
Nothing that happened was publicly reported, so how would you have any idea of what really happened?

Your version of events is wrong in every detail and you should get your facts straight before you decide to condemn others in public. Obviously, you are getting your facts from a biased source.

There is a LOT here that I have kept quiet for a good reason - I don't intend to make any accusations that cannot be absolutely proved and you should do the same.

Besides not knowing the facts, nobody should be criticized for expecting the clearly stated rules to actually be respected. Also, nobody should have to fight to get a rule enforced and it should not even be up to the players or operators which rules they choose to honor or ignore because every result affects every player, not just the two playing.

Although they game ended with a forfeit, there is an interesting side-note. Near the end of the game, a draw by repetition happened - Junior had 2 rooks vs a king and could not checkmate (a common bug that I have had myself.) I was annoyed by this too because if you are in the mode of ignoring rules they might make a case that the game should be considered a win anyway (out of good sportsmanship) and I half expected to see yet another disconnection.
I said what looked like a won game because in the end it was a draw because of a bug. My source of the facts is quite good and I was there when your operator played on.
But there is still much you don't know. For instance I was there from the very beginning and was never consulted on this, and the rules state that the actual programmer need to be on-line too (in the case where there is a non-author operator) to help settle disputes. I do not hold this against Gary the operator, he was just reacting to the situation and he may not have even known I was on-line. Also, I had no knowledge of what he did or said privately to the Junior people.

You also make it sound like I waited an hour and then decided to steal a win but in fact it took a great deal of time to get this resolved, not to mention the fact that the TD took some time to respond to me. I was involved immediately in trying to figure out what happened. I even shouted right away about why Junior was suddenly making moves but was silent about reporting moves and scores and search depths and such. I also asked for a log from Junior but there was no response to that. All of this stuff was going on which you either didn't report or didn't know about.

So your reporting of the incident is full of subjective and emotional language. Example: "43 minutes later when Junior had what looked like a won game the Komodo programmer turned up and claimed the game. "

You wording of the incident is extremely biased and unfair and despite what facts you think you have, you are not reporting them, you are instead just painting a picture and a very unfair picture at that.
User avatar
Harvey Williamson
Posts: 2026
Joined: Sun May 25, 2008 11:12 pm
Location: Whitchurch. Shropshire, UK.
Full name: Harvey Williamson

Re: Looks like Deep Sjeng wins cct12

Post by Harvey Williamson »

Don wrote:
Harvey Williamson wrote:
Don wrote:
Harvey Williamson wrote:
michiguel wrote:
Harvey Williamson wrote:
Albert Silver wrote:
Damir wrote:You forgot to mention it is using 128 cores cluster. On the other hand, It would be a complete failure, if it did not win the tournament...
Considering that huge clusters have more often not won, than won, it is a very nice victory. A hearty congrats to Gian-Carlo, and to Peter Skinner for a great tourney once again!
Sjeng and Rybka are the only top programs that, currently, use Clusters and would have good chances of winning without using a Cluster. Rybka has won CCT the last few years not sure how many were on its Cluster.

I wish we could afford 80+ i7 cores :)

A nice tournament involving MI5 and Mossad. Congrats to Sjeng and well done Peter. The only thing that spoilt it for me was the incident in the last round when Junior was forced to forfeit its game.
What happened? I fell asleep last round.

Miguel
In the final round Junior disconnected for just over the 5 minutes allowed and then returned. The Komodo operator agreed to play on. 43 minutes later when Junior had what looked like a won game the Komodo programmer turned up and claimed the game. The claim was upheld.
Nothing that happened was publicly reported, so how would you have any idea of what really happened?

Your version of events is wrong in every detail and you should get your facts straight before you decide to condemn others in public. Obviously, you are getting your facts from a biased source.

There is a LOT here that I have kept quiet for a good reason - I don't intend to make any accusations that cannot be absolutely proved and you should do the same.

Besides not knowing the facts, nobody should be criticized for expecting the clearly stated rules to actually be respected. Also, nobody should have to fight to get a rule enforced and it should not even be up to the players or operators which rules they choose to honor or ignore because every result affects every player, not just the two playing.

Although they game ended with a forfeit, there is an interesting side-note. Near the end of the game, a draw by repetition happened - Junior had 2 rooks vs a king and could not checkmate (a common bug that I have had myself.) I was annoyed by this too because if you are in the mode of ignoring rules they might make a case that the game should be considered a win anyway (out of good sportsmanship) and I half expected to see yet another disconnection.
I said what looked like a won game because in the end it was a draw because of a bug. My source of the facts is quite good and I was there when your operator played on.
But there is still much you don't know. For instance I was there from the very beginning and was never consulted on this, and the rules state that the actual programmer need to be on-line too (in the case where there is a non-author operator) to help settle disputes. I do not hold this against Gary the operator, he was just reacting to the situation and he may not have even known I was on-line. Also, I had no knowledge of what he did or said privately to the Junior people.

You also make it sound like I waited an hour and then decided to steal a win but in fact it took a great deal of time to get this resolved, not to mention the fact that the TD took some time to respond to me. I was involved immediately in trying to figure out what happened. I even shouted right away about why Junior was suddenly making moves but was silent about reporting moves and scores and search depths and such. I also asked for a log from Junior but there was no response to that. All of this stuff was going on which you either didn't report or didn't know about.

So your reporting of the incident is full of subjective and emotional language. Example: "43 minutes later when Junior had what looked like a won game the Komodo programmer turned up and claimed the game. "

You wording of the incident is extremely biased and unfair and despite what facts you think you have, you are not reporting them, you are instead just painting a picture and a very unfair picture at that.
So why did the TD say:
What Don did was unreal. Totally unsportsmanllike
Highendman

Re: Looks like Deep Sjeng wins cct12

Post by Highendman »

Don wrote:
Harvey Williamson wrote:
Don wrote:
Harvey Williamson wrote:
michiguel wrote:
Harvey Williamson wrote:
Albert Silver wrote:
Damir wrote:You forgot to mention it is using 128 cores cluster. On the other hand, It would be a complete failure, if it did not win the tournament...
Considering that huge clusters have more often not won, than won, it is a very nice victory. A hearty congrats to Gian-Carlo, and to Peter Skinner for a great tourney once again!
Sjeng and Rybka are the only top programs that, currently, use Clusters and would have good chances of winning without using a Cluster. Rybka has won CCT the last few years not sure how many were on its Cluster.

I wish we could afford 80+ i7 cores :)

A nice tournament involving MI5 and Mossad. Congrats to Sjeng and well done Peter. The only thing that spoilt it for me was the incident in the last round when Junior was forced to forfeit its game.
What happened? I fell asleep last round.

Miguel
In the final round Junior disconnected for just over the 5 minutes allowed and then returned. The Komodo operator agreed to play on. 43 minutes later when Junior had what looked like a won game the Komodo programmer turned up and claimed the game. The claim was upheld.
Nothing that happened was publicly reported, so how would you have any idea of what really happened?

Your version of events is wrong in every detail and you should get your facts straight before you decide to condemn others in public. Obviously, you are getting your facts from a biased source.

There is a LOT here that I have kept quiet for a good reason - I don't intend to make any accusations that cannot be absolutely proved and you should do the same.

Besides not knowing the facts, nobody should be criticized for expecting the clearly stated rules to actually be respected. Also, nobody should have to fight to get a rule enforced and it should not even be up to the players or operators which rules they choose to honor or ignore because every result affects every player, not just the two playing.

Although they game ended with a forfeit, there is an interesting side-note. Near the end of the game, a draw by repetition happened - Junior had 2 rooks vs a king and could not checkmate (a common bug that I have had myself.) I was annoyed by this too because if you are in the mode of ignoring rules they might make a case that the game should be considered a win anyway (out of good sportsmanship) and I half expected to see yet another disconnection.
I said what looked like a won game because in the end it was a draw because of a bug. My source of the facts is quite good and I was there when your operator played on.
But there is still much you don't know. For instance I was there from the very beginning and was never consulted on this, and the rules state that the actual programmer need to be on-line too (in the case where there is a non-author operator) to help settle disputes. I do not hold this against Gary the operator, he was just reacting to the situation and he may not have even known I was on-line. Also, I had no knowledge of what he did or said privately to the Junior people.

You also make it sound like I waited an hour and then decided to steal a win but in fact it took a great deal of time to get this resolved, not to mention the fact that the TD took some time to respond to me. I was involved immediately in trying to figure out what happened. I even shouted right away about why Junior was suddenly making moves but was silent about reporting moves and scores and search depths and such. I also asked for a log from Junior but there was no response to that. All of this stuff was going on which you either didn't report or didn't know about.

So your reporting of the incident is full of subjective and emotional language. Example: "43 minutes later when Junior had what looked like a won game the Komodo programmer turned up and claimed the game. "

You wording of the incident is extremely biased and unfair and despite what facts you think you have, you are not reporting them, you are instead just painting a picture and a very unfair picture at that.
Don - you probably don't know me but I have read your interview posted some time ago with a lot of interest and pleasure, and I have the utmost respect to you and GM Kaufman. In addition the emergence of Komodo and the performance it shows on a single core is very promising and exciting for us computer-chess fans. Consider me a very likely future customer when your commercial multi-processors version is out for sale.

I must say though that regarding cct12, from what has been presented so far, your decision to claim a win over Junior disconnect surprised me. Yes, it does seem unsportsmanlike. While you don't owe anyone an explanation, I think the onus is on you to clear the story on this, if you care about the general perception at all, that is.

If you have other facts that you can not disclose in order not to attack other players it's understood. I just wanted to share with you that your decision to claim a win over disconnect (even if the rules allowed that) wasn't something I and I guess many others expected to see. I prefer settling things on the board. If you were worried Junior was cheating post their disconnect (maybe that's what your comment about Junior not reporting scores was about) that's a different argument - but would appear unlikely if Junior wasn't even able to mate a lone king vs. two rooks.

Again, with the utmost respect to your work.

HEM
Zatarra

Re: Looks like Deep Sjeng wins cct12

Post by Zatarra »

"But there is still much you don't know. For instance I was there from the very beginning and was never consulted on this, and the rules state that the actual programmer need to be on-line too (in the case where there is a non-author operator) to help settle disputes. I do not hold this against Gary the operator, he was just reacting to the situation and he may not have even known I was on-line. Also, I had no knowledge of what he did or said privately to the Junior people. "

So in accordance with the methods used to enforce other rules, violation of that rule should have forced forfeiture and removal from the tourney.